REVI EW

EVANDRO AGAZZI and MASSIMO PAURI (eds)

The Reality of the Unobservable: Observability, Unobservability and
Their Impact on the Issue of Scientic Realism
Dordrecht and London: Kluwer Academic Publishers, 2000
cloth $150.00
ISBN 0792363116
Anjan Chakravartty
Department of Philosophy, University of Toronto
There is perhaps no more succinct a way of describing the controversy
between scientic realists and antirealists than to say that it turns on the
reality of the unobservable. Less concisely, it turns on whether we have
reason to think that scientic theories tell us the truth (or something close to
it) about some of the underlying, unobservable bits of a mind-independent,
external reality, among other things. Claims to knowledge of such a reality
have traditionally been a bone of contention between realists and empiricists.
Two decades ago, this ongoing debate was inamed by the introduction of
Bas van Fraassens particular brand of empiricism. Wholesale idealists and
phenomenalists have been increasingly marginalized. The reality of the
observable is now generally taken for granted by most parties to the debate.
The epistemic status of the unobservable, however, remains controversial.
The most recent instalment of the Boston Studies in the Philosophy of
Science series gives the impression that there is very little work to be done on
the issue. The Reality of the Unobservable gives this impression because
practically nothing in it concerns the reality of the unobservableat least not
as it relates to the issue of scientic realism! There is no extended discussion
of the relative epistemic merit of claims about putative observable and
unobservable entities and processes. This is not at all to say that the book is
without merit. But it is only fair that prospective readers be warned that the
title is misleading. Those hoping to nd a signicant concentration of
considered ideas relating to the title subject will be disappointed.
The book is a collection of essays comprising revised versions of papers
presented at a colloquium in Parma in 1995. There are 26 essays plus an
introduction by the editors, and the contents are divided into four main
sections: a general section on realism, one on observation, and two on
quantum theory. The introduction contains 21 pages of detailed summaries
of each of the papers in turna series of de facto abstracts, very handy for
Brit. J. Phil. Sci. 54 (2003), 359363, axg020
&British Society for the Philosophy of Science 2003
previewing particular papers or sections. The quality of the essays is highly
variable, and given the brevity of most, provocative suggestion rather than
developed insight is the norm. Many of the essays, some of which are
intriguing in their own right, have nothing to say about observability or
realism (e.g. Paul Churchland on a proposed model for cognition, Peter
Galison on random number generators). Several do broach these subjects,
but mostly en passant and primarily by recapitulating well-known arguments
(or begging questions against them). A handful have no argued philosophical
conclusions at all. Nevertheless, those attracted by the title of the book may
nd themselves stimulated by threads that run through it, some of which I
will trace here.
In Testability and Empiricism, Dudley Shapere notes that much current
research in particle physics and cosmology concerns things that are not
amenable to observation. In keeping with his earlier work on observation,
Shapere here intends something other than the traditional philosophical
distinction between things that are detectable by the unaided senses and things
that are not. For Shapere, observation occurs whenever information is
transmitted directly (without interference) from the entity in question to an
appropriate receptor. Thus, capturing solar neutrinos here on earth allows us
to directly observe the core of the sun, as opposed to indirect detections of
the stellar core which might be carried out by examining electromagnetic
radiation emanating from the solar surface. Neutrinos generally pass
unaltered from the core to our instruments of detection; information carried
by photons, on the other hand, is profoundly altered along the same journey.
In the context of the book, it is interesting to consider whether Shaperes
distinction between observation and mere detection has any bearing on the
question of realism. That is, we might ask: is it epistemically signicant, or is
it merely a terminological distinction? Shaperes distinction between
observation and mere detection depends, it seems, on a distinction between
direct and indirect evidence. What distinguishes direct from indirect evidence
is the causal history of the information carrier. Neutrinos in the above
example are unaltered, by which I take it that Shapere means they undergo
no causal interactions which change the relevant informational content (the
values of certain of their properties) prior to capture. Their photon
counterparts, conversely, undergo countless interactions which alter the
information they carry.
My suspicion is that these distinctions are not by themselves epistemically
signicant. The reason that they are not has to do with the fact that, if one
believes that we can have information about things not detectable by the
senses, the desired information may be retrievable regardless of whether the
informational content of a carrier has been altered. The reliability of
information retrieval does not depend on there being no prior interfering
360 Review
causal interactions; rather, it depends on the quality of our knowledge of
these interactions. Shapere is quick to emphasize the role of background
information in determining whether or not a detection counts as an
observation. As Brigitte Falkenburg notes in her piece, How to Observe
Quarks, these assumptions must include theories of the entities involved, the
detector, and the transmission of information. What is crucial epistemically, I
suggest, is the quality of our background knowledge. The previous causal
interactions experienced by an information carrier may make retrieving the
desired information computationally more complex, but it does not follow
that this complexity will, in general, prove a greater epistemological challenge
than measuring the properties of an unaltered carrier. Neutrinos, for
example, are particularly elusive.
One of Shaperes motivations is to give an account of the concept of
observation as it functions in the scientic context as opposed to the
philosophical one. What his demarcation between the observable and the
unobservable contributes to the issue of realism is a separate, appropriately
philosophical question. Criteria for warranted claims are, of course, of
central importance to the realist. Recall Ian Hackings inuential position
that realists should be realists not about scientic theories, but about certain
scientic entitiesnamely, those that we can manipulate and use as tools to
intervene in the phenomena. In Measurability, Computability and the
Existence of Theoretical Entities, Mauro Dorato holds that experimental
practice harbours implicit standards for the endorsement of specic,
postulated entities. In addition to manipulabililty, he suggests measurability
as a sucient condition for existence claims regarding unobservables in the
traditional sense (things not detectable by the unaided senses).
Doratos contention is that measuring the values of two or more properties
of a putative entity is sucient to warrant an existence claim. As standards
go, this one is signicantly more permissive than Hackings, for it is easy to
imagine situations in which we might take measurements of things that we are
unable to manipulate or use as tools. This suggests a hierarchy of possible
realist criteria with respect to questions of ontology. Hackings is most
conservative. Anyone holding that we should believe only in those
unobservables that are observable in Shaperes sense would admit a larger
ontology. On Doratos scheme, we would believe in yet more things, on the
assumption that in some cases we are able to measure properties of entities
that are not observable in Shaperes sense. More or less strict criteria, not to
mention variations on the above, are no doubt easy to multiply. Two sorts of
questions spring to mind here from the perspective of the realism debate. The
rst concerns whether any of these pieces of advice for where to draw the line
between what is and is not reasonable to believe engage in a non-question-
begging way with antirealists such as instrumentalists or constructive
The British Journal for the Philosophy of Science 361
empiricists. My guess is that these kinds of speculations engage only the
already committed. In any case, on what epistemic basis should realists prefer
one to another? Some further advice on how to weigh Doratos prescription
against those of others is required.
Of course, being careful to admit only well grounded entities into ones
ontology is not the only way to be a sophisticated realist. In Observation,
Contextuality, and Realism, Bernard dEspagnat takes a dierent tack. He
harks back to the 19th century when (in his estimation) physicists adopted the
wise policy of suspending concern over the fundamental natures of things,
and instead concentrated on giving an account of their behaviours. It is
debatable whether this is a fair historical portrait. DEspagnat cites the
example of Fourier, who gave an equation for the propagation of heat, but
apparently did not take sides on the issue of whether heat is caloric or
molecular agitation. But of course there were physicists who were concerned
with precisely this latter question. Nevertheless, the picture is striking for its
sketch of a position that has returned to contemporary discussions in the
philosophy of science: structural realism. In dEspagnats view, nineteenth-
century scientists could be realists about heat as an entity without worrying
about its nature, for we have an idea or sense of what heat is independently of
any scientic denition, thus obviating the need for one. The same cannot be
said, however, of most of the putative entities of modern physics. We have no
sense of what they are apart from their scientic descriptions.
DEspagnat recommends that we interpret much of science, in keeping with
his nineteenth-century moral, as describing the behaviours of systems, but in
disanalogous fashion, as making no existential claims. Physics makes
existential claims only to the extent that it reveals what cannot exist. To
accept this recommendation is to give up on the idea that science aims at the
description of reality in itself, the real such as it truly is (p. 252). The neo-
Kantian tenor of these words echoes in Charles Paul Enzs contribution,
Observability and Realism in Modern Experiments with Correlated Quantum
Systems. Considering experiments in condensed matter physics, Enz concludes
that we should view the wave function, C, as standing for a real but to an extent
noumenal, unknowable thing. This position is less restrictive than dEspag-
nats, in that Enz does at least countenance the existence of theoretical
entities based on our knowledge of various properties. In his words, [w]e do
not know what the object behind its properties really is (p. 306).
What Enz and dEspagnat have in common is a scepticism about the
natures of the entities that lie behind our observations and detections.
Without attempting to do justice to the current literature on structural
realism, let me strike one note of caution. What is it that the authors above
think we are missing when we characterize entities and processes in terms of
their behaviours or their properties (these may well amount to the same thing,
362 Review
since behavioural relations might be construed as relations between proper-
ties). It is questionable whether there is any principled distinction to be had
between behaviours and properties on the one hand, and natures on the
other. Presumably it is in the nature of heat, for example, to behave in certain
ways. Presumably properties, which are dispositionally responsible for the
phenomena we investigate, are part of the natures of things as well.
DEspagnat and Enz are not merely observing that our knowledge is
incomplete; complete knowledge was never a necessary condition for realism.
What they are suggesting is that the sciences (in particular, physics) give us
one kind of knowledge, but not another. It is unclear, however, that there is
any distinction in kind of the sort they suggest.
The Kantian theme emerges provocatively to pose further questions
elsewhere. Mario Casartelli considers the use of scientic notation in Formal
Representation and the Subjective Side of Scientic Realism, hinting that
notational representations may inuence our views on ontology. This is
certainly possiblenotation may determine what we think it reasonable to
quantify over, and thus what sorts of entities and processes we recognize. But
do our notations condition our understanding of reality, or are they
conditioned by such understanding? Perhaps this is one chicken-and-egg
dispute we need not resolve. A certain amount of taxonomic relativity need
not be an anathema to realism. In Scientic Realism and Scientic
Practice, Roberto Torretti disputes a view that he believes is typical of
modern scientic realism: the idea that reality is well-dened, in some
privileged way, independently of human thought. In the language of Platos
non-vegetarian metaphor of cutting nature at the joints, the idea is that there
is only one correct, objective, natural kind description of the contents of
reality. Torretti advocates something a little more pragmatic, along the lines
of Putnams internal realism.
I am not sure that the sort of realism Torretti contests is, in fact, especially
typical of contemporary scientic realism. It is certainly one option, and there
are realists who believe it. But many I think (including some who dispute
Putnams account) see nothing inconsistent in the idea that knowledge of the
world, including its unobservable elements, can be perfectly objective in the
sense of the realist, yet characterizable in terms of dierent ontological
conventions. The truth is out there, external and mind-independent; we can
know it and cut it up how we like. The challenge for the realist, particularly
with respect to the reality of the unobservable, is in knowing where along the
line to stop, on the continuous spectrum which runs between old-school
realism and full-blown Kantianism.
The British Journal for the Philosophy of Science 363